Crowdsource, Crawl, or Generate? Creating SEA-VL, a Multicultural Vision-Language Dataset for Southeast Asia Paper • 2503.07920 • Published 23 days ago • 95
Intuitive physics understanding emerges from self-supervised pretraining on natural videos Paper • 2502.11831 • Published Feb 17 • 18
Learning Getting-Up Policies for Real-World Humanoid Robots Paper • 2502.12152 • Published Feb 17 • 40